Complications and opioid-prescribing patterns following genicular nerve radiofrequency ablation versus intra-articular injection: a matched cohort study

Abstract

Background and objectives Genicular nerve radiofrequency ablation (GNRFA) is an emerging procedure used to relieve pain from severe knee osteoarthritis. While there have been rare reports of significant complications, their incidence has not been well established. The objective of this study was to examine complication rates at 30 and 90 days post treatment as well as characterize opioid-prescribing patterns before and after treatment. Methods A large national database was queried to identify patients undergoing GNRFA from 2015 to 2022 and matched to control cohorts composed of patients receiving either intra-articular corticosteroid (CSI) or hyaluronic acid injection (HAI) of the knee. Complication rates at 30 and 90 days were analyzed. Opioid utilization was assessed in the 6 months before and after treatment. Results Rates of infection (0.1%, CI 0.02% to 0.5% vs 0.2%, CI 0.05% to 0.6%), septic arthritis (<0.1%, CI 0.003% to 0.4% vs 0.1%, CI 0.02% to 0.5%), deep vein thrombosis (2.0%, CI 1.3% to 2.7% vs 1.6%, CI 1.0% to 2.2%), pulmonary embolism (1.2%, CI 0.6% to 1.7% vs 1.3%, CI 1.1% to 2.5%), bleeding (<0.1%, CI 0.003% to 0.4% vs 0%, CI 0% to 0.3%), seroma (<0.1%, CI 0.003% to 0.4% vs 0.2%, CI 0.05% to 0.6%), nerve injury (0%, CI 0% to 0.3% vs 0%, CI 0% to 0.3%) and thermal injury (0%, CI 0% to 0.3% vs 0%, CI 0% to 0.3%) were not different between CSI and GNRFA cohorts at 30 days. Rate of swelling was significantly greater in the GNRFA cohort (9.4%, CI 7.6% to 10.4% vs 6.4%, CI 4.8% to 7.2%, p=0.003) at 30 days. At 90 days, rates of septic arthritis (0.1%, CI 0.02% to 0.5% vs 0.3%, CI 0.08% to 0.7%), deep vein thrombosis (3.1%, CI 2.1% to 3.8% vs 3.1%, CI 2.2% to 3.9%), pulmonary embolism (1.5%, CI 0.9% to 2.1% vs 1.8%, CI 1.2% to 2.5%), and nerve injury (0%, CI 0% to 0.3% vs <0.1%, CI 0.003% to 0.4%) were not significantly different. Between HAI versus GNRFA cohorts, no significant differences were seen in rates of infection (0.3%, CI 0.08% to 0.07% vs 0.7%, CI 0.3% to 1.2%), septic arthritis (0.2%, CI 0.05% to 0.6% vs 0.4%, CI 0.2% to 0.9%), deep vein thrombosis (2.0%, CI 1.3% to 2.7% vs 1.9%, CI 1.2% to 2.7%), pulmonary embolism (1.5%, CI 0.9% to 2.2% vs 1.7%, CI 1.1% to 2.5%), bleeding (0.1%, CI 0.02% to 0.5% vs 0.2%, CI 0.05% to 0.6%), seroma (<0.1%, CI 0.03% to 0.4% vs 0%, CI 0% to 0.3%), nerve injury (0%, CI 0% to 0.3% vs 0%, CI 0% to 0.3%), swelling (14.0%, CI 11.6% to 15.1% vs 12.0%, CI 10.3% to 13.6%), and thermal injury (<0.1%, CI 0.03% to 0.4% vs <0.01%, CI 0.3% to 0.4%) at 30 days. Rates of infection (0.7%, CI 0.3% to 1.2% vs 1.4%, CI 0.9% to 2.1%), septic arthritis (0.3%, CI 0.1% to 0.8% vs 0.5%, CI 0.2% to 1.1%), deep vein thrombosis (3.6%, CI 2.6% to 4.4% vs 3.1%, CI 2.2% to 4.0%), pulmonary embolism (2.3%, CI 1.5% to 3.0% vs 2.1%, CI 1.4% to 3.0%) and nerve injury (0%, CI 0% to 0.3% vs 0.1%, CI 0.02% to 0.5%) were not significantly different at 90 days. There were no significant differences in level of pretreatment opioid utilization although overall consumption in mean daily morphine equivalents was greater in the GNRFA cohort. Opioid utilization significantly increased in the first 30 days after ablation in patients with no prior opioid use compared to controls. In patients with some and chronic prior opioid use, opioid requirements were generally decreased in all treatment groups at 6 months with no clearly superior treatment in reducing opioid consumption. Conclusion Our study demonstrated that GNRFA possesses a safety profile similar to that of intra-articular injections although significant adverse events such as venous thromboembolism and septic arthritis may occur rarely. Although opioid utilization generally increased in the 30 days after ablation compared with intra-articular injection, similar reduction in opioid consumption at 6 months was seen in patients with prior opioid use in the ablation and control cohorts. Data are available upon reasonable request. All data were obtained via IBM (now Merative) Marketscan Research Databases.